this pdf is a selection from an out-of-print volume from the … · 2020. 3. 20. · 4 navigating...

This PDF is a selection from an out-of-print volume from the NationalBureau of Economic Research

Volume Title: Innovation Policy and the Economy, Volume 1

Volume Author/Editor: Adam B. Jaffe, Josh Lerner and Scott Stern,editors

Volume Publisher: MIT Press

Volume ISBN: 0-262-60041-2

Volume URL: http://www.nber.org/books/jaff01-1

Publication Date: January 2001

Chapter Title: Navigating the Patent Thicket: Cross Licenses, PatentPools, and Standard Setting

Chapter Author: Carl Shapiro

Chapter URL: http://www.nber.org/chapters/c10778

Chapter pages in book: (p. 119 - 150)

4

Navigating the Patent Thicket: Cross Licenses,Patent Pools, and Standard Setting

Carl Shapiro, University of Ca4fornia at Berkeley

Executive Summary

In several key industries, including semiconductors, biotechnology, computersoftware, and the Internet, our patent system is creating a patent thicket: anoverlapping set of patent rights requiring that those seeking to commercializenew technology obtain licenses from multiple patentees. The patent thicket isespecially thorny when combined with the risk of holdup, namely the dangerthat new products will inadvertently infringe on patents issued after theseproducts were designed. The need to navigate the patent thicket and holdup isespecially pronounced in industries such as telecommunications and comput-ing in which formal standard setting is a core part of bringing new technologiesto market. Cross licenses and patent pools are two natural and effective meth-ods used by market participants to cut through the patent thicket, but each in-volves some transaction costs. Antitrust law and enforcement, with itshistorical hostility to cooperation among horizontal rivals, can easily add tothese transaction costs. Yet a few relatively simple principles, such as the desir-ability package licensing for complementary patents but not for substitute pat-ents, can go a long way toward insuring that antitrust will help solve theproblems caused by the patent thicket and by holdup rather than exacerbatingthem.

I. The Patent Thicket

Is our patent system slowing down the commercialization of newtechnologies?

The essence of science is cumulative investigation combined withhypothesis testing. The notion of cumulative innovation, each discov-ery building on many previous findings, is central to the scientificmethod. Indeed, no respectable scientist would fail to recognize andacknowledge the crucial role played by his or her predecessors in es-tablishing a foundation from which progress could be made. As Sir

Shapiro120

Isaac Newton put it, each scientist "stands on the shoulders of giants"

to reach new heights.Today, most basic and applied researchers are effectively standing on

top of a huge pyramid, not just on one set of shoulders. Of course, apyramid can rise to far greater heights than could any one person, es-pecially if the foundation is strong and broad. But what happens if, in

order to scale the pyramid and place a new block on the top, a research-

er must gain the permission of each person who previously placed ablock in the pyramid, perhaps paying a royalty or tax to gain such per-mission? Would this system of intellectual property rights slow downthe construction of the pyramid or limit its height?

Clearly, pyramid building, namely research and development(R&D), is taking place at an impressive pace today, so there is no great

cause for alarm, especially in the area of basic research where the "roy-alty" is often (but not always) nothing more than a citation. As we

move from pure R to applied R and ultimately to D, however, one canfairly ask whether our legal and commercial institutions are in factproperly designed to promote rather than discourage the creation ofproducts and services that draw on many strands of innovation andthus potentially require licenses from multiple patent holders. To com-

plete the analogy, blocking patents play the role of the pyramid's build-

ing blocks.Mixing metaphors, thoughtful observers are increasingly expressing

concerns that our patent (and copyright) system is in fact creating apatent thicket, a dense web of overlapping intellectual property rightsthat a company must hack its way through in order to actually com-mercialize new technology. With cumulative innovation and multipleblocking patents, stronger patent rights can have the perverse effect of

stifling, not encouraging, innovation.1In fact, even while a consensus has emerged that innovation is the

main driver of economic growth, we are witnessing somewhat of abacklash against the patent system as it is currently operating. Espe-

cially unpopular are patents on business methods, such asPriceline.com's patent on "buyer-driven conditional purchase offers"(asserted against Microsoft) or Amazon's patent on a one click online

shopping system (asserted against Barnes & Noble). The Patent andTrademark Office (PTO) does indeed seem to have allowed a number

of patents on ideas that would not appear offhand to meet the usualstandards for novelty and nonobviousness, such as the patent held bySightsound.com which reputedly covers "the sale of audio or video re-

Navigating the Patent Thicket121

cordings in download fashion over the Internet." Emboldened by a keyappeals court decision in 1998 supporting a patent for a businessmethod enabled by computer software, patent applications for com-puter-related business methods have jumped from about 1,000 in 1997to over 2,500 in 1999. In an attempt to call a truce in what could other-wise prove to be a mutually destructive patent battle, Jeff Bezos, theChairman of Amazon.com, recently suggested that patents on softwareand Internet business methods be limited to 3 or 5 years, rather thanthe usual 20 years from the date of application.2

But concerns about a patent thicket, and excessively loose standardsat the PTO, are hardly confined to e-commerce and business methodpatents. For example, in the semiconductor industry, companies suchas IBM, Intel, or Motorola find it all tooeasy to unintentionally infringeon a patent in designing a microprocessor, potentially exposing them-selves to billions of dollars of liability and/or an injunction forcingthem to cease production of key products.3 So-called submarine pat-ents, that take years if not decades to work their way through the Pat-ent and Trademark Office, are another great source of anxietyespecially for large manufacturing firms. Plus, more and more compa-nies are following the lead of Texas Instruments and engaging in patentmining, trying to get the most out of their patents by asserting themmore aggressively than ever against possible infringing firms, eventhose who are not rivals. And considerable research shows that compa-nies are increasingly inclined to seek patents, causing an increase in thepropensity to patent, as well as an increase in the practice of defensivepatenting.4

In short, our patent system, while surely a spur to innovation over-all, is in danger of imposing an unnecessary drag on innovation by en-abling multiple rights owners to "tax" new products, processes, andeven business methods. The vast number of patents currently being is-sued creates a very real danger that a single product or service will in-fringe on many patents. Worse yet, many patents cover products orprocesses already being widely used when the patent is issued, makingit harder for the companies actually building businesses and manufac-turing products to invent around these patents. Add in the fact that apatent holder can seek injunctive relief, that is, can threaten to shutdown the operations of the infringing company, and the possibility forholdup becomes all too real.

This paper takes as given the flood of patents currently being issuedby the PTO, and assumes that these patents are indeed creating a

122Shapiro

patent thicket in the sense that many new products would likely in-

fringe on multiple patents. Remaining agnostic (but suspicious) about

whether the PTO is too lax in granting patents (especially software pat-

ents), or whether the courts are too generous in upholding patents that

are granted, I look at the business arrangements that are being used to

cut through the patent thicket.More specifically, I consider the evolving and growing role of cross li-

censes and patent pools to solve the complements problem that arises

when multiple patent holders can potentially block a given product. I

discuss specifically the standard setting process, that increasingly in-

volves complex negotiations over patent rights and licensing terms. I

also consider other ways in which companies resolve disputes overin-

tellectual property, including acquisitions.For each business practice, in addition to describing the economics

underlying that practice and examples of its use, I consider whether

antitrust limits are contributing to the problems caused by the patent

system. Unfortunately, antitrust enforcement and antitrust law have a

deep rooted suspicion of cooperative activities involving direct com-

petitors. But such cooperation1 in one form or another, may be precisely

what is required to navigate the patent thicket. As a result, unless anti-

trust law and enforcement are quite sensitive to the problemsposed by

the patent thicket, they can have the perverse effect of slowing down

the commercialization of new discoveries and ultimately retarding in-

novation, precisely the opposite of the intent of both the patent laws

and the antitrust laws.

II. Market Responses to Overlapping Patents

The Economic Theory of Complements

The generic problem inherent in the patent thicket is well understood

as a matter of economic theory, at least in its static version. Consider,

for example, a company seeking to manufacture a new graphics chip

for use in personal computers or video game consoles. (Substitute a

biotech firm using patented tools for genetic engineering, or an e-com-

merce firm using patented business methods, if you would prefer.)

Suppose that the company's preferred design for this chip is likely to

infringe on a number of patents; the process manufacturing methods

used to actually produce the chip infringe on a number of additional


patents. In order to produce the chip as designed, the company needsto obtain licenses from a number, call it N, of separate rights holders.

This situation is precisely the classic complements problem origi-nally studied by Cournot in 1838. Cournot considered the problemfaced by a manufacturer of brass who had to purchase two key inputs,copper and zinc, each controlled by a monopolist.5 As Cournot demon-strated, the resulting price of brass was higher than would arise if a sin-gle firm controlled trade in both copper and zinc, and sold these inputsto a competitive brass industry (or made the brass itself). Worse yet, thecombined profits of the producers were lower as well in the presence ofcomplementary monopolies. So, the sad result of the balkanized rightsto copper and zinc was to harm both consumers and producers.6 Thesame applies today when multiple companies control blocking patentsfor a particular product, process, or business method.

How can the inefficiency associated with multiple blocking patentsbe eliminated? One natural and attractive solution is for the copper andzinc suppliers to join forces and offer their inputs for a single, packageprice to the brass industry. The two monopolist suppliers will find it intheir joint interest to offer a package price that is less than these two com-ponents sold for when priced separately. The blocking patent version ofthis principle is that the rights holders will find it attractive to create apackage license or patent pool, or in some situations to simply engagein cross licensing so they can each produce final products themselves.

The appendix offers a short, modern, and more general version ofCournot's theory of complements cast in terms of blocking patents.This basic theory of complements (used in fixed proportions) givesstrong support for businesses to adopt, and for competition authoritiesto welcome, either cross licensees, package licenses, or patent pools toclear such blocking positions. If two patent holders are the only compa-nies realistically capable of manufacturing products that utilize theirintellectual property rights, a royalty-free cross license is ideal from thepoint of view of ex post competition. But any cross license is superior toa world in which the patents holders fail to cooperate, since neithercould proceed with actual production and sale in that world withoutinfringing on the other's patents. Alternatively, if the two patent hold-ers see benefits from enabling many others to make products that uti-lize their intellectual property rights, a patent pool, under which all theblocking patents are licensed in a coordinated fashion as a package, canbe an ideal outcome. The simple theory, which is sketched out in the

124Shapiro

appendix, suggests that coordinating such licensing can lead to lower

royalty rates than would independent pricing (licensing) of the two

companies' patents.In other words, without cross licenses or patent pools, there is a ten-

dency for products to bear multiple patent burdens. The buildup of li-

censing fees can have several unattractive consequences. First, the

well-known costs of static monopoly power are magnified: prices are

well above marginal costs, causing inefficiently low use of these prod-

ucts. As shown in the appendix, with N rights holders, equilibrium

markups are N times the monopoly level. Of course, this is merely a

magnified version of the monopoly burden resulting from the patent

system itself, but it is well to remember Cournot's lesson that the multi-

ple burdens reduce both consumer welfare and the profits of patentees

in comparison with a coordinated licensing approach. Second, these

burdens may cause certain products not to be produced at all, if that

production is subject to economies of scale. Third (this is a dynamic

version of the previous point), the prospect of paying such royaltiesnecessarily reduces the return to new product design and develop-

ment, and thus can easily be a drag on innovation and commercializa-

tion of new technologies.Heller and Eisenberg (1998) discuss the complements problem in the

context of biotechnology patents, making a nice comparison to the clas-

sic tragedy of the commons. The well-known tragedy of the commons

refers to the fact that a resource can be overused if it is not protected by

property rights; fishing grounds and clean water are standard exam-

ples. Heller and Eisenberg point out that quite a different problemarises when there are multiple blocking patents; they label this problem

the tragedy of the anti-commons. The tragedy of the anti-commons

arises when there are multiple gatekeepers, each of whom must grant

permission before a resource can be used. With such excessive property

rights, the resource is likely to be underused. In the case of patents, in-

novation is stifled.

The Holdup Problem

As noted above, the complements problem is at its worst when thedownstream firms using the various inputs truly require each input to

make their products. In the patent context, if a manufacturer finds itrelatively easy to design around a given patent, the royalties that the

patentee can assert are necessarily limited. So, unless the patent in


question is quite broad, one might think that any burden on the manu-facturer would be modest, and arguably the very return we wish toprovide to the patentee as a reward for innovation.

Unfortunately, this rather romantic view of patents is less and lessapplicable in our economy, for three reasons. First, even a modest tax iscounterproductive if the patent was improperly granted, that is, if thepatentee did not truly made a new and useful discovery, or if the patentas granted was too broad, covering some prior art as well as somethingtruly new. Second, the cumulative effect of many small taxes can be-come quite large; there are sound reasons to believe that the staticdeadweight loss associated with these royalties is increasing and con-vex in the tax rate, at least over some range of royalties. The danger ofpaying royalties to multiple patent owners is hardly a theoretical curi-osity in industries such as semiconductors in which many thousands ofpatents are issued each year and manufacturers can potentially in-fringe on hundreds of patents with a single product.

Third, and most important, is timing. Suppose that our repre-sentative manufacturer could, with ease, invent around a given patent,if that manufacturer were aware of the patent and afforded sufficientlead time. Clearly, in this case the patented technology contributes littleif anything to the final product, and any reasonable royalty would bemodest at best. But, oh, how the situation changes if the manufacturerhas already designed its product and placed it into large scale produc-tion before the patent issues. In this case, even though the timing isstrongly suggestive that the manufacturer did not in fact rely on thepatented invention for the design of its product, the manufacturer is ina far weaker negotiating position. The patentee can credibly seek fargreater royalties, very likely backed up with the threat of shuttingdown the manufacturer if the Court indeed finds the patent valid andinfringed and grants injunctive relief. The manufacturer could go backand redesign its product, but to do so (a) could well require a major re-design effort and/or cause a significant disruption to production, (b)would still leave potential liability for any products sold after the pat-ent issued before the redesigned products are available for sale, and (c)could present compatibility problems with other products or betweendifferent versions of this product. In other words, for all of these rea-sons, the manufacturer is highly susceptible to holdup by the patentee.I submit that this holdup problem is very real today, and that both pat-ent and antitrust policymakers should regard holdup as a problem offirst order significance in the years ahead.

126Shapiro

The holdup problem is worst in industries where hundreds if not

thousands of patents, some already issued, others pending, can poten-

tially read on a given product. In these industries, the danger that a

manufacturer will step on a land mine is all too real. The result will be

that some companies avoid the mine field altogether, that is, refrain

from introducing certain products for fear of holdup. Other companies

will lose their corporate legs, that is, will be forced to pay royalties on

patents that they could easily have invented around at an earlier stage,

had they merely been aware that such a patent either existed or was

pending. Of course, ultimately the expected value of these royalties

must be reflected in the price of final goods.In short, with multiple overlapping patents, and under a system in

which patent applications are secret and patents slow to issue (relative

to the speed of new product introduction), we have a volatile mix of

two powerful types of transaction costs that can burden innovation: (1)

the complements problem, the solution of which requirescoordination,

perhaps large scale coordination; and (2) the holdup problem, which is

quite resistant to solution in the absence of either (a) better information

at an earlier stage about patents likely to issue, and/or (b) the abilityof

interested parties to challenge patents at the PTO before they have is-

sued and are given some presumption of validity by the Courts.

Clearly, these concerns form the basis for a serious discussion about

reform of the patent system.7 However, my intention in this paper is to

explore how private companies canbest navigate the patent system we

currently have, and how our antitrust laws can be enforced in a way

that is sensitive to the transaction costs associated with our current pat-

ent system. I see relatively little that private companies can do to over-

come the holdup problem without reform of the patent system itself.

But there is quite a bit they can do to solve the complements problem,

which itself is greatly exacerbated by the holdup problem.

Overlapping Patents and Business Strategy in Practice

To solve the complements problem generally, and to cut through the

patent thicket specifically, requires coordination among rights holders.

Such coordination itself faces two types of obstacles. First, there are in-

evitably coordination costs that must be overcome. Second, antitrust

sensitivities are invariably heightened when companies in the same or

related lines of business combine their assets, jointly set fees of any

sort, or even talk directly with one another. Because such coordination


may involve the elimination of competition, we have a complex inter-action between private and public interests. Even as coordination be-tween rights holders is critical, from a public policy perspective wecannot presume that private deals are in the public interest. Antitrustauthorities will legitimately want to know whether consumers arehelped or harmed by any arrangement; injured parties may seek re-dress under the antitrust laws or by alleging patent misuse.

Cross Licenses Cross licenses commonly are negotiated when each oftwo companies has patents that may read on the other's products orprocesses. Rather than blocking each other and going to court or ceas-ing production, the two enter into a cross license. Especially with a roy-alty free cross license, each firm is then free to compete, both indesigning its products without fear of infringement and in pricing itsproducts without the burden of a per unit royalty due to the other.Thus, cross licenses can solve the complements problem, at leastamong two firms, and thus be highly procompetitive.

A cross license is simply an agreement between two companies thatgrants each the right to practice the other's patents. Cross licenses mayor may not involve fixed fees or running royalties; running royaltiescan in principle run in one direction or both. Cross licenses may in-volve various field-of-use restrictions or geographic restrictions. Crosslicenses may involve some but not all relevant patents held by eitherparty; carve-outs are not uncommon. And cross licenses, like regular li-censes, may be confined to patents issued (or pending) as of the date ofthe license, or they may include patents to be granted through a certaintime in the future.

Patent Pools and Package Licenses When two or more companies con-trol patents necessary to make a given product, and when at least someactual or potential manufacturers may not themselves hold any suchpatents, a patent pool or a package license can be the natural solutionto the complements problem. Under a patent pool, an entire group ofpatents is licensed in a package, either by one of the patent holders orby a new entity established for this purpose, usually to anyone willingto pay the associated royalties. Under a package license, two or morepatent holders agree to the terms on which they will jointly licensetheir complementary patents and divide up the proceeds. A nice exam-ple of a patent pool is the Manufacturers Aircraft Association formedin 1917 to license a number of patents necessary for the production of

Shapiro128

airplanes, patents controlled by The Wright-Martin Aircraft Corpora-

tion, the Curtiss Aeroplane & Motor Corporation, and others.8 I discuss

below some more recent patent pools that have been used to help es-

tablish compatibility standards.

Cooperative Standard Setting The need to solve the complements prob-

lem tends to be especially great in the context of standard setting. For

example, when the International Telecommunications Union (ITU) es-

tablishes a new standard for fax transmissions or modem protocols, the

participants are loath to agree to a standard that can be controlled by

any single firm through its patents. Thus, standard setting organiza-

tions like the ITU or the American National Standards Institute (ANSI)

typically require that participants agree to license all patents essential

to compliance with any standard on "fair, reasonable, and nondiscrimi-

natory" terms. Rules such as this are explicitly intended to reduce or

eliminate any holdup problems. However, it is well to note that many

standard setting organizations are wary of sanctioning any specific

agreement regarding the magnitude of licensing terms for fear of anti-

trust liability, as such agreements might be construed as price fixing.

Perversely, by leaving the precise licensing terms vague, this caution

can in fact lead to ex post holdup by particular rights holders, contrary

both to the goal of enabling innovation and to consumers' interests.

The case in which multiple firms control patents essential to a stan-

dard fits well with the formal economic analysis described above. In es-

sence, any manufacturer seeking to produce a compliant product must

obtain a license from each rights holder to avoid facing an infringe-

ment action. Inventing around is typically impractical, as it would pre-

clude the manufacturer from claiming that its products are compliant

and thus assuring consumers that they are fully compatible with theprevailing standard. Thus, standard setting very often has especially

strong elements of both the complements problem and the holdup

problem.

Settlements of Patent Disputes Cross licenses (or simply licenses) are a

common way in which companies resolve patent disputes. But other

forms of settlement arise, two of which I touch on below. First, I discuss

acquisitions, in which one firm simply acquires the other, thereby re-

solving the dispute and assembling the various intellectual property

rights within a single company. Second, I comment on cash payments

in exchange for exit, a strategy whereby one company pays the other

company to exit the market, and thus to drop its challenge to the first


company's patent. In each of these cases, legitimate questions arise asto whether any particular private agreement truly is in the publicinterest.

Antitrust Limits

As I have indicated, many of the business solutions to the complementsproblem and the holdup problem raise antitrust issues. Quite generally,agreements among companies that either do compete, or might com-pete, directly with each other raise antitrust warning flags. For eachbusiness form, I consider below its antitrust treatment.

Generally speaking, one can imagine two rather different ap-proaches that antitrust might take to firms' efforts to coordinate tosolve the complements problem. One approach is to ask whether theagreement in question leads to more competition than would occurwithout that agreement. This is the approach advocated in the Depart-ment of Justice and Federal Trade Commission Antitrust Guidelines forthe Licensing of Intellectual Property, which state in §3.1 that:

However, antitrust concerns may arise when a licensing arrangement harmscompetition among entities that would have been actual or likely potentialcompetitors in a relevant market in the absence of the license (entities in a "hor-izontal relationship").

Another quite different approach would be to ask whether the agree-ment in question is the most competitive agreement possible. Put dif-ferently, one could ask whether a given agreement is the leastrestrictive alternative that is workable in the sense of solving the legiti-mate business problem faced, such as unbiocking patent positions.Clearly, this latter standard, which does not reflect current antitrust en-forcement policy according to the Guidelines, would be far tougher onall forms of cooperation among patent and copyright holders.

III. Cross Licensing

Cross Licenses and Design Freedom

Cross licenses are the preferred means by which large companies clearblocking patent positions amongst themselves. Based in part on work Ihave done on behalf of Intel, I can report that broad cross licenses arethe norm in markets for the design and manufacture of microproces-sors.9 For example, Intel has entered into a number of broad cross

130Shapiro

licenses with other major industry participants, such as IBM, under

which most of each company's vast patent portfolio is licensed to the

other. Furthermore, the companies generally agree to grant licenses to

each other for patents that will be issued several years into the future,

typically for the lifetime of the cross licensing agreement. Often, these

cross licenses involve no running royalties, although they may involve

balancing payments at the outset to reflect differences in the strength of

the two companies' patent portfolios as reflected in a patentpageant,

and/or the vulnerability of each to an infringement action by the other.

For example, Hewlett-Packard and Xerox recently announced a cross

license that settled their outstanding patent disputes.

From the perspective of competition policy, cross licenses of this sort

are quite attractive. The traditional concern with cross licenses among

competitors is that running royalties will be used as a device to elevate

prices and effect a cartel; see Katz and Shapiro 1985. Clearly, such con-

cerns do not apply to licenses that involve small or no running royal-

ties, but rather have fixed up-front payments. Another concern is that

the granting of licenses to future patents will reduce each company's in-

centive to innovate because its rival will be able to imitate its improve-

ments.10 While correct in theory it is clear, at least in the case of

semiconductors and no doubt more widely, that this concern is

dwarfed by the benefits arising when each firm enjoys enhanced de-

sign freedom by virtue of its access to the other firm's patent portfolio.

There is little doubt that these broad cross licenses permit the more

efficient use of engineers (arguably the resource that governs the rate of

innovation in the semiconductor industry), better products, and faster

product design cycles. In other words, when IBM and Intel sign a for-

ward looking cross license, each is enabled to innovate more quickly

and more effectively without fear that the other will hold it up by as-

serting a patent that it has unintentionally infringed. And neither firm

is really all that worried that the other will actually copy its products,

just because the other has a license to most of its patents. Of course, the

impressive rate of innovation in the semiconductor industry in the

presence of a web of such cross licenses offers direct empirical support

for the view that these cross licenses promote rather than stifle

innovation.

Intel's Policy of "IPfor IP"

Despite all of these benefits, the Federal Trade Commission attacked

Intel's cross licensing practices in 1998.11 One key episode behind the


FTC's complaint involved Intel's conduct when faced with a lawsuit byIntergraph, a workstation manufacturer, asserting that Intel's micro-processors infringed on certain patents held by Intergraph. Of course,lawsuits like Intergraph's are a necessary part of the threat point be-hind any cross-licensing negotiation: if one party is not happy with theterms offered by the other, it always has the option of initiating patentlitigation. In response to Intergraph's infringement action against Intel,Intel withdrew its own intellectual property from Intergraph by suingIntergraph for infringement of Intel's patents and by withdrawing thesupply of Intel trade secrets to Intergraph, trade secrets that Intergraphvalued highly for the purposes of designing systems built on Intelchips.

Evidently viewing Intel's conduct as unfair, the FTC attempted tofashion an antitrust case against Intel based on this conduct, along witha similar response by Intel to a lawsuit initiated by Digital EquipmentCorporation.12 The FTC action against Intel sharply exposed the factthat the FTC and Intel had fundamentally different views about the im-pact of the conduct at issue. The FTC saw Intel as using its existing mo-nopoly power to fortify its position by lowering its royalty costs perchip and potentially offering superior products by incorporating tech-nologies patented by others. Intel viewed itself as engaging in a defen-sive exercise which was a necessary aspect of cross licensing, namelytrading intellectual property for intellectual property (IP for IP) andwithdrawing its own intellectual property when faced with a frontalassault on its core product line in the form of an infringement actionseeking injunctive relief. Intel, well aware of what a juicy target itposed, believed it had every right to protect itself from holdup, andcertainly no duty to give special treatment in the form of Intel trade se-crets and advance product samples to a company attempting to holdit up.

The problem for the FTC was that the conduct at issue, especiallywith respect to Intergraph, was directed at a customer of Intel's, not acompetitor. Brushing aside concerns about holdup, and playing downthe important role of cross licenses in the semiconductor industry, theFTC found no "business justification" for Intel's conduct, and thus wasprepared to presume that the conduct was anticompetitive without ac-tually studying the impact of the conduct on Intel's competitors. Infact, Intel's true rivals in microprocessor design and manufacturing(such as AMD, Motorola, Sun, or IBM) were either not subject to theconduct at issue (since they were not Intel customers at all and thus notrecipients of the Intel trade secrets at issue), or had ongoing cross

132Shapiro

licenses with Intel under which the litigation triggering these episodes

would simply not occur in the first place.Fortunately, a compromise was reached and a settlement agreed to

between the FTC and Intel.'3 In essence, Intel agreed not to withdraw

product information needed by its customers to build systems based

on soon-to-be-released Intel chips. (Presumably, thispromise provides

some benefit to Intel by assuring its customers that they will not be held

up once they are relying on Intel for their newsystems.) But Intel is not

obligated to continue to provide trade secrets on products farther out

on their roadmap (i.e., products that will not be introduced for a year

or two) to customers suing Intel, and Intel was notobligated to provide

any trade secrets to a company suing Intel and seeking a court injunc-

tion to shut down Intel's microprocessor business.

The Intel situation also exposes the interplay between government

enforcement of the antitrust laws and private antitrust actions. Even

while the FTC was investigating Intel, bringing a complaint against

Intel, and ultimately settling with Intel, Intergraph was engaged in its

own antitrust and patent battle with Intel. Intergraph won a resound-

ing victory in the first round of that battle, in which the District Court

judge in Alabama issued a searing anti-Intel opinion ruling, among

other things, that Intel's microprocessors and associated trade secrets

were "essential facilities" under antitrust laws, thus imposing a duty

on Intel to sell its microprocessors to Intergraph and to make its trade

secrets available to Intergraph, Intergraph's lawsuit against Intel not-

withstanding. This opinion was based on strands of antitrust law that

require dominant companies to deal with their rivals, especially if the

dominant firm has established an ongoing course of dealing with rivals

in the past.14Ultimately, however, Intel was vindicated. The District Court judge

later ruled that Intel was not in fact infringing on Intergraph's patents.

And, most significantly, the Court of Appeals for the Federal Circuit

vacated the District Court's antitrust and essential facility opinion.'5 In

a strongly worded and sweeping opinion, theappeals court ruled that

Intel's conduct did not violate the antitrust laws because it was not di-

rected at a competitor and indeed could have no adverse impact on

competition in the market where Intel was alleged to have monopoly

power, namely the market for microprocessors1 in which Intergraph

did not compete. The FTC's efforts to fashion an antitrust case out of

Intel's conduct look even more dubious now in the light of this subse-

quent decision by the Court of Appeals.


The Intel episode is closely related to another ongoing debate regard-ing the intersection between intellectual property rights and anti-trust law: can a company violate the antitrust laws simply by refusingto license its patents, or by refusing to sell patented items, to its ri-vals? Most commentators have said for some time that a refusal to li-cense patents cannot in and of itself constitute an antitrust violation.However, the Supreme Court has signaled that unilateral refusalsto sell can indeed constitute antitrust violations, especially if a com-pany has established an ongoing course of dealing with its rivals.16The precise conditions under which a refusal to license a patent(or to sell patented items) could constitute an antitrust violationhas remained unclear. Most observers were stunned when the NinthCircuit Court of Appeals ruled in 1997 that Kodak was liable forrefusing to sell patented spare parts for its machines to independentservice organizations seeking to compete against Kodak in thebusiness of servicing Kodak copiers and micrographics equipment.As the Court acknowledged, this was the first time a unilateralrefusal to sell a patented item had been judged to be an antitrust viola-tion.17 Just recently, the Court of Appeals for the Federal Circuit cameto a very different conclusion, ruling that a company's unilateral deci-sion not to license a patent (or sell a patented item) could never in andof itself constitute an antitrust violation.18 Hopefully, the SupremeCourt will resolve this significant split among the Circuit Courts andclarify that unilateral refusals to license patents are immune from anti-trust challenge.

Intel's practices, and those of other firms who require grantbacks ofrelevant patents in exchange for a license to key enabling patents,copyrights, or trade secrets, raises further interesting questions aboutthe role of self help in the digital economy.19 One view of such businessstrategies cum legal regimes is that they are a welcome effort by lead-ing firms to establish a type of litigation-free zone likely to favor inno-vation and get around some of the current difficulties with our patentsystem and the patent thicket it causes. A less favorable view is thatthese arrangements represent efforts by powerful firms to establish pri-vate legal regimes that favor themselves and make it more difficult forupstarts to challenge the dominance of current market leaders. Is across licensing policy of IP for IP a beneficial way to cut through thepatent thicket, or a strong-arm tactic by a dominant firm that enjoyspowerful patent rights and seeks access to others' intellectual propertyin exchange?

134Shapiro

IV. Patent Pools

A patent pool involves a single entity (either a new entity or oneof the

original patent holders) that licenses the patents of two or more compa-

nies to third parties as a package. In many respects, a patent pool

(much like a package license) is the purest solution to the complements

problem described above and analyzed in the appendix. Indeed, licens-

ees may well welcome such a pool, both for the convenience of

one-stop shopping and because a subset of the required patents may be

of little or no value by themselves. Thus, from the licensee's perspec-

tive, licensing the entire package is simpler and avoids the danger of

paying for some patent rights that turn out to be useless without other

complementary rights.

Essential Patents vs. Rival Patents

The Department of Justice (DOJ) has clearly articulated its policy to-

ward patent pools/package licensing in a trio of businessreview letters

regarding an MPEG patent pool and two DVD patent pools. The es-

sence of this approach which precisely mirrors the economic princi-

ples articulated above, is that inclusion of truly complementary patents

in a patent pool is desirable and procompetitive but assembly of substi-

tute or rival patents in a pool can eliminate competition and lead to ele-

vated license fees. But differently, the key distinction in forming a

patent pool is that between blocking or essential patents, which properly

belong in the pool, and substitute or rival patents, which may need to re-

main separate.In the MPEG case,2° the Department approved the creation of a pool

of patents necessary to enable manufacturers to meet the MPEG-2

video compression technology. This pool, encompassing patentsfrom

Fujitsu, General Instrument, Lucent, Matsushita, Mitsubishi, Philips,

Scientific-Atlanta, Sony, and Columbia University, permits one-stop

shopping for makers of televisions, digital video disks and players,and

telecommunications equipment as well as cable, satellite, and broad-

cast television services. To support their formation of a patent pool,

these nine patent holders conducted an extensive search toidentify all

patents essential to the MPEG-2 standard and include them in the pool.

The licensing agent for the pool, MPEG LA, will employ an independ-

ent patent expert to determine whether a patent in the pool is in fact


essential, and whether other patents as well are essential and thus suit-able for inclusion in the pool. As stated by the Department, "the use ofthe independent-expert mechanism will help ensure that the portfoliowill contain only patents that are truly essential to the MPEG-2 stan-dard, weeding out patents that are competitive alternatives to eachother."

In the first Digital Versatile Disk (DVD) case,21 the Department ap-proved a proposal by Philips, Sony, and Pioneer to jointly license pat-ents necessary to make discs and players that comply with theDVD-Video and DVD-ROM standards. Again, only essential patentsare to be included in the joint licensing program. As with the earlier CDlicensing program of Sony and Philips, licenses will be offered byPhilips, in this case on behalf of all three firms. Again, an independentpatent expert will be employed to ensure that the license only conveysthe rights to essential patents. As stated by the Department, "the expertwill help ensure that the patent pool does not combine patents thatwould otherwise be competing with each other." The Department sub-sequently approved a second joint licensing scheme relating to theD\TDVideo and DVD-ROM standards,22 this one including patentsheld by Toshiba (the licensing entity), Hitachi, Matsushita, Mitsubishi,Time Warner, and Victor Company of Japan. Note that the effect ofthese two patent pools appears to be to reduce but not eliminate thecomplements problem, since there remain two separate pools, not justone: two-stop shopping, it would appear.

A Patent Pool Created to Resolve Claims of Blocking Patents

In contrast to the Department of Justice's approval of these three patentpools, the Federal Trade Commission in March 1998 challenged a pat-ent pooi formed by Summit Technology, Inc. and VisX, Inc., two firmsthat manufacture and market lasers to perform a new, and increasinglypopular, vision correcting eye surgery, photorefractive keratectomy.23According to the FTC: "Instead of competing with each other, the firmsplaced their competing patents in a patent pool and share the proceedseach and every time a Summit or VISX laser is used." The FTC was os-tensibly following the same principles employed by the Justice Depart-ment, namely to permit the assembly of complementary or essentialpatents, but not rival patents, into a pool. According to the FTC,the two companies agreed not to license their patents independently.

Shapiro136

However, the companies in this case argued vigorously that they did

indeed have mutually blocking patents, making their pool, Pillar Point

Partners, procompetitive. In August 1998 the two companies settledwith the FTC and agreed to lift any restrictions on each other regarding

the licensing of their patents; ultimately, their patent pool was

dissolved.24The Summit and VisX case raises a number of very interesting and

tricky issues regarding patent pools and joint licensing programs in

general. First, if two companies reasonably believed that their patents

blocked each other at the time they formed the pool, was that sufficient

to justify the formation of a pool? How hard are they required to look

into the validity of each other's claims before agreeing to pool their

patents? Second, if each firm believed it could, at considerable expense,

delay, and risk, invent around the other's patents, should the two firms

be prohibited from forming a pool and rather forced to attempt to in-

vent around each other's patents, under the view that consumers might

thereby enjoy the benefits of direct competition (although the product

might be delayed, or never introduced, in the absence of the pool)?

Third, is there competitive harm in placing some potentially rival pat-

ents into the pool, assuming that each party in fact controls valid block-

ing patents, making some type of pool procompetitive? Fourth, can the

pool be attacked on antitrust grounds based on the argument that a less

restrictive alternative, namely a cross license, would have achieved the

same legitimate purposes and created additional ex post competition? If

so, does it matter in this assessment if the two companies agree that the

pool will license their patents to third parties, something that a cross

license would not permit, unless it contained rather unusual sub-

licensing rights?

V. Cooperative Standard Setting

Blocking patents are especially common in the context of standard set-

ting: once a standard is picked, any patents (or copyrights) necessary to

comply with that standard become truly essential. If the standard be-

comes popular, each such patent can confer significant market power

on its owner, and the standard itself is subject to holdup if these patent

holders are not somehow obligated to license their patents on reason-

able terms. As noted above, for precisely this reason, standard settingbodies require participants to license any essential patents on reason-

able terms as a quid pro quo before adopting any standards.25


Fortunately, antitrust concerns have not prevented a great many co-operative standard setting efforts from proceeding forward. Some par-ticipants go so far as to say that much of the innovation taking placenow in the telecommunications, Internet, and computer areas is stan-dards based. Indeed, even the fiercest enemies often team up in thesoftware industry to promote new standards. Back in 1997, Microsoftand Netscape, two companies hardly known as cozy partners, agreedto include compatible versions of Virtual Reality Modeling Language(developed by Silicon Graphics) in their browsers. This agreement wasexpected to make it far easier for consumers to view 3D images on theWeb. Earlier, Microsoft agreed to support the Open Profiling Standard,which permits users of personal computers to control what personalinformation is disclosed to a particular web site, and which had previ-ously been advanced by Netscape, along with Firefly Network, Inc.and Verisign Inc.

But neither is cooperative standard setting immune from antitrustscrutiny. In the consumer electronics area, for example, the Justice De-partment investigated Sony, Philips, and others regarding the estab-lishment of the CD standard in the 1980s. Cooperative efforts to setoptical disc standards have also been challenged in private antitrustcases, on the theory that agreements to adhere to a standard are an un-reasonable restraint of trade:

Edlefendants have agreed, combined, and conspired to eliminate competition. by agreeing not to compete in the design of formats for compact discs and

compact disc players, and by instead agreeing to establish, and establishing, acommon format and design.

Does cooperation lead to efficient standardization, increased compe-tition, and additional consumer benefits? Or is cooperative standardsetting a means for firms collectively to stifle competition, to the detri-ment of consumers and firms not included in the standard settinggroup? Answering these questions and evaluating the limits thatshould be placed on cooperative standard setting efforts require ananalysis of the competitive effects of such cooperation in comparisonwith some reasonable but-for world. Inevitably, an antitrust analysis ofcooperative standard setting involves an assessment of how the marketwould likely evolve without the cooperation. One possibility is thatmultiple, incompatible products would prevail in the market, if not forthe cooperation. Another possibility is that the market would eventu-ally tip to a single product, even without cooperation. Even in this

138Shapiro

latter case, an initial industrywide standard can have significant

efficiency and welfare consequences, for three reasons: (1) cooperation

may lock in a different product design than would emerge from com-

petition; (2) cooperation may eliminate a standards war waged prior to

tipping; and (3) cooperation is likely to enable multiple firms to supply

the industry standard product, whereas a standards war may lead to a

single, proprietary product.

The Costs and Benefits of Compatibility and Standards

There are significant benefits associated with achieving compatibility

These include:

successful launching of a bandwagon or network,

greater realization of network effects,

protecting buyers from stranding, and

enabling competition within an open standard.

Likewise, standardization and compatibility can impose very real

costs on consumers:

constraints on variety and innovation,

loss of ex ante competition to win the market, and

proprietary control over a closed standard.

Legal Treatment of Cooperative Standard Setting

I now look more closely at the intellectual property issues that arise

specifically in the context of standard setting, where the participants

typically agree to license their patents on fair, reasonable, and nondis-

criminatory terms.Firms are sometimes accused of hiding intellectual property rights

until after the proprietary technology has been embedded in a formal

standard. I view this issue primarily as one of contract law. Standard

setting groups typically have provisions in their charters compelling

participants either to reveal all relevant intellectual property rights or

to commit to licensing any intellectual property rights embedded in the

standard on reasonable terms.27 Clearly these rules help control the

holdup problem. In some cases, however, the precise requirements im-

posed by a standard setting group may be unclear. In these circum-

stances, if the standard affects nonparticipants, including consumers,


there is a public interest in clarifying the duties imposed on partici-pants in a fashion that promotes rather than stifles competition.

The question of whether firms should be allowed, or even encour-aged, to set standards cooperatively is part of the broader issue of col-laboration among competitors, a storied area within antitrust law. Mostof the case law deals with quality and performance standards ratherthan compatibility standards.28 Existing cases also have tended to focuson the standard setting process itself, rather than the outcomes of coop-erative standard setting.

Antitrust liability has been found for participants in a standard set-ting process who abuse that process to exclude competitors from themarket. One leading case is Allied Tube & Conduit Corp. v. Indian Head,Inc., 486 U.S. 492 (1988), in which the Supreme Court affirmed a juryverdict against a group of manufacturers of steel conduit for electricalcable. These manufacturers conspired to block an amendment of theNational Electric Code that would have permitted the use of plasticconduit. They achieved this by packing the annual meeting of the Na-tional Fire Protection Association, whose model code is widelyadopted by state and local governments. The other leading case isAmerican Society of Mechanical Engineers v. Hydrolevel Corp., 456 U.S. 556(1982), in which the Supreme Court affirmed an antitrust judgmentagainst a trade association. In this case, the chairman of an associationsubcommittee offered an unofficial ruling that the plaintiff's productwas unsafe, and this ruling was used by the plaintiff's rival (who en-joyed representation on the subcommittee) to discourage customersfrom buying the plaintiff's product.

Antitrust risks associated with excluding a rival from the market ap-pear to be less of a problem for an open standard, but could arise ifthe companies promoting the standard block others from adheringto the standard or seek royalties from outsiders. The DOJ business re-view letters regarding the MPEG-2, DVD-Video, and DVD-ROM stan-dards are excellent illustrations of how the enforcement agencies cansuccessfully handle intellectual property in the standard settingcontext.

As the Supreme Court has noted, "Agreement on a product standardis, after all, implicitly an agreement not to manufacture, distribute, orpurchase certain types of products."29 To date, this type of reasoninghas not been used to impose per se liability on software standard set-ting activities. Indeed, I know of no successful antitrust challengesto cooperation to set compatibility standards. The closest case of whichI am aware is Addamax Corporation v. Open Software Foundation, Inc.,

140Shapiro

888 F. Supp. 274 (1995). In Addamax, the District Court refused to

grant summary judgment on behalf of the Open Software Foundation,

an industry consortium formed to develop a platform-indepen-

dent version of the UNIX operating system. OSF conducted a bid-

ding to select a supplier of security software. After failing to be se-

lected, Addamax brought antitrust claims against OSF, Hewlett-

Packard, and Digital Equipment Corporation, asserting that OSF

had chosen the winner not based on the merits but to favor specific

companies and technologies. The Addamax case looks problematic,

inasmuch as the primary purpose of OSF was to permit its members

to team up to offer stronger competition against the leadingUNIX

vendors, Sun Microsystems and AT&T, and there was no evidence

suggesting that OSF's failure to pick Addamax was based on its

members desire to control the market in which Addamax itself

operated.Ultimately, the antitrust risks faced by companies that are trying to

set compatibility standards appear to be relativelyminor as long as the

scope of the agreement truly is limited to standard setting and steers

clear of distribution, marketing, and pricing. While the lawhas typi-

cally looked for integration and risk-sharing amongcollaborators in or-

der to classify cooperation as a joint venture and escape per secondemnation, these are not very helpful screens for standard setting

activities. The essence of cooperative standard, setting is not the sharing

of risks associated with specific investments, or theintegration of oper-

ations, but rather the contribution of complementary intellectual prop-

erty rights and the expression of unified support to ignite positive

feedback for a new technology.The limits imposed by public policy in the area of compatibility stan-

dards remain unclear. The most specific statement by the antitrust en-

forcement agencies can be found in a recent FTC Staff Report.3° The

Staff Report recognized a need for clarification in this area:

the time has come for a significant effort to rationalize, simplify and articulate

in one document the antitrust standards that federal enforcerswill apply in as-

sessing collaborations among competitors. This effort should bedirected at

drafting and promulgating "competitor collaboration guidelines"that would

be applicable to a wide variety of industry settings and flexibleenough to ap-

ply sensibly as industries continue rapidly to innovate and evolve.3'

Since that call for action, the FTC has conducted Joint Venture

Hearings, and the Commission and the Antitrust Division issued in


April 2000 new "Antitrust Guidelines for Collaborations Among Com-petitors" (available at either Agency's web site).

Hidden Patents and Holdup in Standard Setting

A number of disputes have surfaced recently that illustrate the thornyproblems associated with hidden patent rights that were later exertedagainst established standards.32

Dell Computer and the VESA VL-Bus Standard The leading U.S. exam-ple of this type of antitrust action is the FTC's consent agreement withDell Computer Corporation, announced in November 1995. Althoughthe case involved computer hardware, it is important for the softwarecommunity as well. The assertion was that Dell threatened to exerciseundisclosed patent rights against computer companies adopting theVL-bus standard, a mechanism to transfer data instructions betweenthe computer's CPU and its peripherals such as the hard disk drive orthe display screen. The VL-bus was used in 486 chips, but it has nowbeen supplanted by the PCI bus. According to the FTC.

During the standard-setting process, VESA [Video Electronics Standard Asso-ciationj asked its members to certify whether they had any patents, trade-marks, or copyrights that conflicted with the proposed VL-bus standard; Dellcertified that it had no such intellectual property rights. After VESA adoptedthe standardbased in part, on Dell's certificationDell sought to enforce itspatent against firms planning to follow the standard.33

There are two controversial issues surrounding this consent decree:(a) the FTC did not assert that Dell acquired market power, and indeedthe VL-bus never was successful; and (b) the FTC did not assert thatDell intentionally misled VESA. My analysis suggests thatanticompetitive harm is unlikely to arise in the absence of significantmarket power and that the competitive effects are not dependent onDell's intentions.

Motorola and the IT1I V34 Modem Standard Another good example ofhow competition can be affected when standard setting organizationsimpose ambiguous duties on participants is the case of Motorola andthe V.34 modem standard adopted by the International Telecommuni-cations Union. Motorola agreed to license its patents essential to thestandard case to all comers on "fair, reasonable, and nondiscriminatory

142Shapiro

terms."34 Once the standard was in place, Motorola then made offers

that some industry participants did not regard as meetingthis obliga-

tion. Litigation ensued between Rockwell andMotorola, in part over

the question of whether reasonable terms should mean:(a) the terms

that Motorola could have obtained ex ante, in competitionwith other

technology that could have been placed in the standard; or (b) the

terms that Motorola could extract ex post, giventhat the standard is set

and Motorola's patents are essential to that standard.

These issues are best dealt with by the standard setting bodies, or

standard setting participants either by making more explicitthe duties

imposed on participants1 or by encouraging ex ante competition

among different holders of intellectual propertyrights to get their

property into the standard. Unfortunately, antitrust concernshave led

at least some of these bodies to steer clear of such exante competition,

on the grounds that their job is merely to settechnical standards, not to

get involved in prices, including the terms on which intellectual prop-

erty will be made available to other participants.The ironic result has

been to embolden some companies to seek substantialroyalties after

participating in formal standard setting activities.

VI. Settlements of Patent Disputes

Cross licenses and patent pools can be ways to settleintellectual prop-

erty disputes. For example, the Summit and VisX patentpool discussed

above, Pillar Point Partners, was essentially a settlement of a patent

dispute between Summit and VisX.Generally speaking, antitrust authorities have legitimate concerns

that parties will settle their intellectual property disputes in waysthat

stifle competition. As a matter of economic theory there is no reasonto

expect the two parties' collective interests in settlement,and especially

in the form of any settlement they adopt, to coincidewith the public in-

terest, which includes consumer interests. So, while the law surely wel-

comes the settlement of disputes generally,and does not seek to force

parties to litigate to the death, some settlements canbe anticompetitiVe.

Based on this general view, Assistant Attorney General for Antitrust

Joel Klein recently suggested (see Klein 1997) thatparties notify the

Justice Department of certain settlements that they enter into,much as

parties are required to notify the Justice Department and the FTC in ad-

vance of their intention to merge.


Firms are quite creative in crafting settlements of intellectual prop-erty disputes, and by no means restrict their attention to cross licensesand patent pools. For example, one tried and true method of settling adispute is for the companies involved simply to merge. However, theantitrust authorities are well aware that such mergers can themselveseliminate competition, and they will view such mergers with skepti-cism if there is a good chance that the two parties will in fact be capableof competing against each other, their patent claims notwithstanding.A good example of such a merger that was modified in response to FTCconcerns was the proposed merger of Boston Scientific and CVIS in thearea of imaging catheters.5 An interesting twist in such cases is that theparties' posturing in court, where they each have an incentive to assertthat they are not infringing on the other's patents, provides direct am-munition to the FTC or DOJ to assert that the two companies could in-deed compete independently if not for the merger.

A second method that companies can use to settle a patent dispute isfor one company to simply pay the other company to drop its claimsand exit the market. Such agreements raise obvious antitrust concerns,because an incumbent firm may be willing to pay handsomely to elimi-nate a potential competitor and avoid the risk of having its patent chal-lenged, especially if no equally effective challenger is likely to arrive onthe scene any time soon. The losers in such deals can easily be subse-quent would-be entrants (if the patent were struck down) or consum-ers (who would benefit from a finding that the patent at issue is invalidor not infringed). Put differently, a settlement can generate negative ex-ternalities, either to other firms or to consumers, and thus there is a le-gitimate role of the Courts and the antitrust enforcement agencies tooversee such settlements.

One class of settlements that are suspicious on their face is that in-volving agreements between incumbent manufacturers of brandedpharmaceuticals and would-be rivals who seek to offer generic compe-tition by challenging the validity of the patents underlying the brandedproduct's dominant position. It has been reported recently that theFTC is considering challenging several such settlements.36 These caseshave an interesting twist resulting from the fact that certain genericmanufacturers can gain preferential rights to enter the market beforeothers are permitted to do so. As a result, the branded manufacturermay be able to stall competition by entering into a suitable agree-ment with the uniquely-placed generic manufacturer, knowing that

144

subsequent rivals will face some delay. In order to identify and

prevent any anticompetitive agreements of this nature,the FTC has

asked that the FDA require companies to notify the FDAof any such

settlements and make that information available to the FTC for its

review.

vii. Conclusions

Our current patent system is causing a potentially dangerous situation

in several fields, including biotechnology,semiconductors, computer

software, and e-commerce, in which a would-be entrepreneuror inno-

vator may face a barrage of infringement actions that it must overcome

to bring its product or service to market. In otherwords, we are in dan-

ger of creating significant transaction costsfor those seeking to com-

mercialize new technology based on multiple patents,overlapping

rights, and holdup problems. Under these circumstances,it is fair to

ask whether the pendulum has swung too far in the direction of

strong patent rights, ranging from the standards used at the Patent and

Trademark Office for approving patent applications1 to the secrecy

of such app1ication5 to the presumption affordedby the courts to pat-

ent validity to the right of patent holders toseek injunctive relief by in-

sisting that infringing firms cease production of the offending

products.Under these circumstances, we can ill afford to further raise transac-

tion costs by making it difficult for patentees possessing complemen-

tary and potentially blocking patents to coordinateto engage in cross

licensing, package licensing, or to form patent pools. Yet antitrust law

can potentially play such a counterproductiverole, especially since an-

titrust jurisprudence starts with a hostility towardcooperation among

horizontal rivals.So far, the Department of Justice has displayed a

keen understanding

of the need for those holding complementaryrights to coordinate in

the licensing of those rights, but the FederalTrade Commission has ex-

hibited less restraint, and arguably is making it moredifficult for firms

to engage in cross licenses, to offer packagelicenses, or to form

procompetitiVe patent pools. Many of these issues are likely to be ex-

tremely important in the near future, especially with the rise of stan-

dard setting as an essential part of the process by which new

technologies are commercialized.

Shapiro


Notes

Prepared for presentation at "Innovation Policy and the Economy," National Bureau ofEconomic Research, Adam Jaffe, Joshua Lerner, and Scott Stern, organizers, April 11,2000, Washington DC. Comments are welcomed; please direct any comments [email protected].

For example, in 1995 Joseph Stiglitz, then Chairman of the Council of Economic Advi-sors, stated at the opening of the Federal Trade Commission's hearings on CompetitionPolicy in the New High-Tech, Global Marketplace, that "some people jump . . . to the con-clusion that the broader the patent rights are, the better it is for innovation, and that isn'talways correct, because we have an innovation system in which one innovation builds onanother. If you get monopoly rights down at the bottom, you may stifle competition thatuses those patents later on and so. . . the breadth and utilization of patent rights can beused not only to stifle competition, but also have adverse effects in the long run on inno-vation." See FFC Staff Report, p. 6.

See http://www.amazoncom/exec/obidos/subst/mi/tthl/lO5496631.

Nearly 5,000 patents were granted in the U.S. in a recent single year, 1998, relating tomicroprocessors alone, not to mention semiconductors more broadly.

See, for example, Kortum and Lerner 1998, Cohen et al. 2000, and Hall and Ham 1999.For a brief description of Cournot's original work on complements, and modernextensions, see Shapiro 1989, p. 339.

Cournot assumed that the two inputs, copper and zinc, were required in certain fixedproportions for the production of brass. Ifone input can be substituted for the other, theyhave properties of substitutes as well as complements, in which case competition betweenthe two input owners can go far to solving the problem posed here. Throughout this pa-per, I am assuming that the company in question requires rights to practice each of sev-eral patents, and that one patent license cannot substitute for another. Clearly, to the ex-tent that a manufacturer, for example, can rely on multiple designs or productionprocesses covered by separate patents with separate owners, the patent thicket is far lessof a problem. But even in this relatively friendly setting, extra difficulties can still beraised by the holdup problem, discussed below.

For a thoughtful discussion of possible reforms at the Patent and Trademark Office,see Merges 1999.

See Klein 1997 for a further description of this pool and how it operated. In this case,the Assistant Secretary of the Navy, Franklin D. Roosevelt, had to leanon the industry toform a pool and help enable wartime production of aircraft.

See Hall and Ham 1999 and Grindley and Teece 1997 for additional studies of licens-ing practices in the semiconductor industry.

This concern about discouraging innovation also arises with respect to grantbacks,under which one company agrees to license its future patents in exchange for rights touse an existing patent held by another company. See Gilbert and Shapiro 1997 for a fur-ther discussion of grantbacks.

In the matter of Intel Corporation, Docket No. 9288, Complaint filed June 8, 1998. The

146Shapiro

Complaint is available athttp://wwW.ftC.gOv/0s/1998/9806/mteffmncmPtm I was

retained by Intel to work on this matter.

For one well-informed articulation of the theory underlying the FTC'sposition, see

Baker 1999.

For more information on the settlement between the FTC and Intel, see

http:/ /ftc.gov/os/1999/9903/d09288mtelagementhtm

The key recent Supreme Court case here is Aspen Skiing Company v. AspenHighlands

Skiing Corp., 472 U.S. 585 (1985), although the essential facilities doctrine goesback to the

case of U.S. v. Terminal Railroad Association of St. Louis, 224 U.S. 383(1912).

Intergraph Corporation v. Intel Corporation,United States Court of Appeals for theFed-

eral Circuit, 98-1308, Decided November 5, 1999, Judge Newman writing theopinion for

the Court.

The classic cites are Otter Tail Power Co. v. U.S., 410 U.S. 366 (1973) (duty to sellwhole-

sale electric power to a retail competitor) and Aspen Skiing Company v. AspenHighlands

Skiing Corp., 472 U.S. 585 (1985), (duty to continue to offer a joint lift ticket with arival ski

slope).

The Court set up a tortured standard under which a company's decision torefuse to

license its patent was "presumptively valid," but could be overcome byevidence that the

company's intent was anticompetitive. Of course, asserting intellectual propertyrights

against a would-be rival is typically anticompetitive in the sense of trying toeliminate a

competitor (or at least earn royalties from the competitor, which add to thecompetitor's

costs), so this test is not in fact workable. Amazingly, the Court said that Kodak wouldbe

justified in refusing to sell patented parts if its intent was to earn a return on itsR&D in-

vestment required to design and manufacture those parts, but not if its intent wasto

eliminate competitors who rely on those verypatented parts. I testified on behalf of Ko-

dak in this case.

United States Court of Appeals for the Federal Circuit, 99-1323, In Re Independent

Service Organizations Antitrust Litigation, CSU, et. al. v. Xerox Corporation,Decided Febru-

ary 17, 2000, Judge Mayer writing the opinion.

For a discussion of self help focusing on copyright holders, see Dam 1998.

See the June 26, 1997 press release athttp://www.usdOj/g0v/atr/Pud/P155_

releases/1997/1173.htm.

See the December 17, 1998 press release at http://www.u5doj.gOv/afr/Pd/

press_releases/1998/2l20htm.

See the June 10, 1999 business review letter at http://wwW.u5dOj.g0v/atr/Pd/pressjeleases/1999/2484.htm

See the March 24, 1998 press release at http://www.ftc.gOv/opa/1998/98031'

eye.htm.

For a description of the settlement, see the August 21, 1998 press release at

http: / /www.ftc.gov/opa/1998/9808/5umw5x.htm.Despite this settlement, the FTC

continued to pursue VisX for allegedly acquiring a key patent by inequitableconduct

and fraud by omission on the U.S. Patent and Trademark Office. However, anadminis-

trative law judge subsequently dismissed this complaint; see the June 4, 1999 press re-

lease at http:/ /www.ftc.gov/opa/1999/9906/w5tm.

Navigating the Patent Thicket 147

Note that these rules can create the perverse incentive for patent holders to assertthat at least some of their patents are not in fact essential, but perhaps merely extremelyhelpful, in complying with the standard. By this device, a patent holdercan in principleeither refuse to license its patent to others (especially once the standard has become es-tablished, and perhaps for a patent that issued after the standard is established) or seeksomething more than fair and reasonable royalties. Of course, whether the terms fair andreasonable are evaluated on an ex ante or ex post basis is not precisely clear, although theterms would have little force if applied only on an ex post basis.

"Second Amended Complaint," Disctronics Texas, Inc., et al. v. Pioneer Electronic Corp.et al. Eastern District of Texas, Case No. 4:95 CV 229, filed August 2, 1996 at 12.

Note that a company might profit from refusing to participate in the standard settingprocess, in the hope that the resulting standard will nonetheless (perhaps inadvertently)infringe on the company's patent. Then the company would not be obligated to licenseits blocking patent on fair and reasonable terms, if at all. This would at least create thepossibility that the company in question could control the standard and make it propri-etary once it became established.

See Anton and Yao 1995 for a more complete discussion of the legal treatment ofper-formance standards.

Allied Tube & Conduit Corp. v. Indian Head, Inc., 486 U.S. 492,500 (1988).

Federal Trade Commission. 1996, June. "Anticipating the 21st Century: CompetitionPolicy in the New High-Tech Global Marketplace," Chapter 9, "Networks and Stan-dards."

ibid, Chapter 10, "Joint Ventures," at 17.

There are many more examples of disputes involving hidden patent rights and stan-dard setting, including: Wang vs. Mitsubishi; Microsoft and Cascading Style Sheets; andETSI and Third-Generation Mobile Telephones.

See http://www.ftc.gov/opa/9606/dell2.htm.

I served as an expert in this matter retained by Rockwell; the views stated here do notnecessarily reflect those of any party to the case.

See the May 3, 1995 press release at http://www.ftc.gov/opa/1995/9505/boscvis.htm. The recent merger of Gemstar and TV Guide is another example of amerger/settlement that raises antitrust issues.

One episode under investigation involves Abbott Laboratories, Novartis's GenevaPharmaceuticals unit, and the popular hypertension drug, Hytrin. Another episode in-volves Aventis (the new company formed from the merger of Hoechst andRhone-Poulenc), Andrx, and the heart drug Cardizem CD. Abbott reportedly agreed topay Geneva $4.5 million per month to delay the launch of a generic version of Hytrin.Abbott asserts that its agreement with Geneva is "in accordance with all laws." See theWall Street Journal, February 7, 2000, "FTC Panel Backs Suit Against Abbott, Novartis onDeal for Hypertension Drug," p. B20. See the FTC website for updates.

References

Anton, James, and Dennis Yao. 1995. "Standard-Setting Consortia, Antitrust, andHigh-Technology Industries." Antitrust Law Journal 64:247-65.

148 Shapiro

Baker, Jonathan B. 1999. "Promoting Innovation Competition Through the Aspen/KodakRule." George Mason Law Review 7:495-521.

Balto, David. 1999. "Networks and Exclusivity: Antitrust Analysis to Promote NetworkCompetition." George Mason Law Review 7:523-76.

Cohen, Wesley M., Richard R. Nelson, and John Walsh. 2000, February. "Protecting TheirIntellectual Assets: Appropriability Conditions and Why U.S. Manufacturing Firms Pat-ent (or Not)." NBER Working Paper No. W7552.

Dam, Kenneth. 1998, August. "Self-Help in the Digital Jungle." John M. Olin Law & Eco-nomics Working Paper no. 59, University of Chicago Law School, Chicago.

Farrell, Joseph and Michael Katz. 1998. "The Effects of Antitrust and Intellectual PropertyLaw on Compatibility and Innovation." Antitrust Bulletin.

Federal Trade Commission. 1996, May. "Competition Policy in the New High-TechGlobal Marketplace." Washington DC. Staff Report.

Gilbert, Richard, and Carl Shapiro. 1997. "Antitrust Issues in the Licensing of IntellectualProperty: The Nine No-No's Meet the Nineties." Brookings Papers on Economics:Microeconomics: 283-336.

Grindley, Peter, and David J. Teece. 1997. "Managing Intellectual Capital: Licensing andCross-Licensing in Semiconductors and Electronics." California Management Review39(2):1-34.

Hall, Bronwyn, and Rose Marie Ham. 1999. "The Patent Paradox Revisited: Determi-nants of Patenting in the U.S. Semiconductor Industry, 1980-94." NBER Working Paper

No. W7062.

Heller, M. A., and R. S. Eisenberg. 1998. "Can Patents Deter Innovation? TheAnticommons in Biomedical Research." Science 280:698-701.

Katz, Michael, and Carl Shapiro. 1985, Winter. "On the Licensing of Innovations." RandJournal of Economics.

Katz, Michael, and Carl Shapiro. 1994. "Systems Competition and Network Effects."Journal of Economic Perspectives 8(2):93-115.

Klein, Joel I. 1997. "Cross-Licensing and Antitrust Law." Available athttp:/ /www.usdoj.gov/atr/public/speeches/1123.htm.

Kortum, S., and J. Lerner. 1998. "Stronger Protection or Technological Revolution: Whatis Behind the Recent Surge in Patenting?" Carnegie-Rochester Conference Series on Pub-lic Policy, Vol. 48 (June 1998): 247-304.

Lemley, Mark, and David McGowan. 1998. "Legal Implications of Network Economic Ef-fects." California Law Review 86:481-611.

Merges, Robert P. 1999. "As Many as Six Impossible Patents Before Breakfast: PropertyRights for Business Concepts and Patent System Reform." Berkeley Technology Law Journal14(2):577-615.

Shapiro, Carl. 1989. "Theories of Oligopoly Behavior." In R. Schmalensee and R. Willig,eds., Handbook of Industrial Organization. New York: Elsevier Science Publishers: 330-414.

Shapiro, Carl. 1996a. "Antitrust in Network Industries." Available athttp: / /www.usdoj/gov/atr/public/speeches/shaPir.mar.

Navigating the Patent Thicket 149

Shapiro, Carl. 1999. "Exclusivity in Network Industries.7:673-84.

Shapiro, Carl, and Hal R. Varian. 1998. Information Rules: AEconomy. Cambridge, MA: Harvard Business School Press.

U.S. Department of Justice and Federal Trade Commission.lines for the Licensing of Intellectual Property. Washington DC.

tJ.S. Department of Justice and Federal Trade Commission.lines for Collaborations Among Competitors. Washington, DC.

Technical Appendix

Here I show that prices can be well above monopoly levels if multiplefirms have critical patents, all of which read on a single product. Moreprecisely, if N firms each control a patent that is essential for the pro-duction of a given product, and if these N firms independently set theirlicensing fees, the resulting markup on that product is N times the mo-nopoly markup.

Suppose that N firms, i = 1,. . . , N, each own a patent that is essentialto the production of a given product. For simplicity let us think of therebeing a competitive industry that produces this product, buying andassembling the necessary components from each of these N firms. Forthis purpose we can think of firm i either as setting a license fee for theuse of its patent, or as setting a price at which it will sell its essentialcomponent to the competitive assembly industry; the theory is identi-cal either way.

The cost to firm i per unit (for making and selling its component orfor licensing its patent to assemblers) is denoted by c. The price ofcomponent i (or the license fee charged by firm i) is denoted by p. Theprice of the product itself is denoted by p. In addition to paying royal-ties (or buying components), the assembly firms incur an assemblycost per unit eual to a. Competition at the "assembly" level ensuresthat p = a + L=1 p.

Demand for the product in question is denoted by D(p). The absolutevalue of the elasticity of demand is given by E = D'(p)p/D(p). In gen-eral, will vary with p.

I assume that the N firms set their component prices, equivalentlytheir license fees, independently and noncooperatively. In other words,I look for the Nash Equilibrium in the prices ri,. . . PN. The profits forfirm i are given by

George Mason Law Review

Strategic Guide to the Network

1995, April. Antitrust Guide-

2000, April. Antitrust Guide-

150Shapiro

'rr = D(p)(p1 - c1).

The first-order condition for firm i is given by

=D(p)+D'(p)(p c)=O.

Adding up across all i gives

D(p)N+D'(p)(p c1)=O.

which can be rewritten as

N ' c ' D(I''' '= "' N.p pD'(p)

Using the definition of the elasticity of demand, and the fact that

p we have

p - (a+

c)N (1)

In other words, the percentage markup over cost for the product inquestion is equal to N times the inverse of the elasticity of demand. In

contrast, the standard monopoly markup rule would be

p - (a±

c)(2)

The markup with N independent firms controlling key patents isequal to N times the monopoly markup.

It can be shown that the combined profits of the N firms under inde-

pendent pricing is lower than would be earned by a monopolist sellingall N components. This implies that the firms have an incentive to coor-

dinate their pricing. A package license for all N components wouldlead to higher (combined) profits and lower prices for consumers.

this pdf is a selection from an out-of-print volume from the … · 2020. 3. 20. · 4 navigating...

Documents